--- title: "Google Cloud Storage Integration" description: "Manage Google Cloud Storage buckets and objects from FlowRunner agents: list, upload, download, copy, and delete, plus V4 signed URLs for time-limited sharing and direct uploads." url: https://flowrunner.ai/integrations/google-cloud-storage date_modified: 2026-08-07T17:20:23-07:00 --- # Google Cloud Storage [Storage](https://flowrunner.ai/integrations/category/storage) Manage Google Cloud Storage buckets and objects from FlowRunner agents: list, upload, download, copy, and delete, plus V4 signed URLs for time-limited sharing and direct uploads. [Verified](https://flowrunner.ai/integrations/verified "What does verified mean?") · 12 actions · Custom auth · available [Google Cloud Storage website](https://cloud.google.com/) · [Platform Documentation](https://cloud.google.com/storage/docs/json_api) · Capability data verified 2026-08-12 1. A flow produces a report or export that needs durable storage 2. Agent reads the file it generated into the flow 3. Agent uploads the object to a bucket from a FlowRunner or public URL 4. Agent updates the object metadata with content type and record tags 5. Agent confirms the object with Get Object Metadata before sharing 6. Agent assembles the object name, recipient, and proposed link scope 7. A person approves the signed URL before an external party can download the file ## What This Integration Enables The Google Cloud Storage connector covers buckets, objects, and signed URLs. Agents list buckets with prefix filtering, read bucket metadata, create buckets with a fixed location and storage class, and delete empty buckets. On objects, agents list with prefix filtering and folder-style browsing, read object metadata, download into FlowRunner file storage, and upload from a FlowRunner or public URL. They delete objects, copy server-side to another name or bucket, and update mutable metadata such as content type and cache control. Generate Signed URL creates a V4 link that grants time-limited GET or PUT access to a single object without Google credentials, up to a 7-day expiration, signed locally with the service account's private key. Some of these actions read, some destroy, and one exposes a file; the connector supplies all of them, and the surrounding flow decides which need a person. ### Without FlowRunner **Exports left in place**: Reports pile up in a working location with no durable home **Sharing by broad access**: Handing a file to a partner means widening access or emailing the raw file **Metadata drifts**: Content type and cache headers get set inconsistently across objects ### With FlowRunner **Exports archived durably**: Every generated file lands in a bucket as part of the flow **Sharing by scoped link**: A partner gets a time-limited signed URL to one object, gated for review **Metadata set in the flow**: Content type, cache control, and custom tags are written with the object ## Use Case Scenarios ### Report Archive and Delivery At month-end, a finance agent generates a set of reports and uploads each one to a bucket under a dated prefix with Upload Object. For the reports a stakeholder needs directly, the agent proposes a signed URL, waits for approval, then calls Generate Signed URL and emails the time-limited link. The reports live in a consistent location, and nobody emails a spreadsheet around. ### Fetch, Process, Re-file A raw document lands in a bucket. The agent calls Download Object to pull it into FlowRunner file storage, runs it through an extraction step, and writes the structured result back with Upload Object under a processed prefix. Copy Object and Delete Object then reorganize the bucket so raw and processed files stay cleanly separated. ### Bulk Reorganization An agent reads a list of object names from a source of truth, then calls Copy Object and Delete Object to move files into a new prefix layout. Because copies run server-side, the data never leaves Google and the reorganization is fast even across large objects. ## Human-in-Loop Highlight Listing buckets, reading metadata, and archiving objects are safe operations a flow runs on its own. Generating a signed URL is different: it produces a working link to a private object that anyone holding it can use, with no Google credentials required, until it expires. This is [human-in-the-loop](https://flowrunner.ai/concepts/human-in-the-loop) at the boundary between storing a file and handing it out. Before generating a signed URL for an external recipient, the agent pauses, assembles the object name, the access mode, the expiration, and the recipient, and routes to a person through their preferred channel. The message is specific: "This would create a 7-day download link to reports/2026/q3-board-pack.pdf for an external board member. Approve the link and its expiration, or shorten it?" The agent handles the upload, the metadata, and the link mechanics; a person owns every link that reaches outside the account. Agent processes routinely Detects exception requiring judgment Clear match Continues automatically Ambiguous Routes to human via preferred channel Human decides Agent resumes with decision ## Agent Capabilities 12 actions ### Buckets 4 - **List Buckets** Lists the buckets in the project with name, location, storage class, and timestamps, filterable by name prefix and paged. - **Get Bucket** Retrieves a bucket's metadata, including location, storage class, versioning state, and labels. - **Create Bucket** Creates a new bucket with a fixed location and default storage class. Bucket names are globally unique across Google Cloud and cannot be changed later. - **Delete Bucket** Permanently deletes an empty bucket. The operation cannot be undone, and it fails if the bucket still contains objects. ### Objects 7 - **List Objects** Lists the objects in a bucket with name, size, content type, and timestamps, supporting prefix filtering and folder-style browsing via a delimiter. - **Get Object Metadata** Retrieves an object's metadata, including size, content type, checksums, cache control, and custom metadata, without downloading its data. - **Download Object** Downloads an object into FlowRunner file storage and returns a URL for later flow steps. - **Upload Object** Uploads a file from a FlowRunner file URL or any public URL into a bucket, inferring content type from the object name when not specified. - **Delete Object** Permanently deletes an object. It cannot be undone unless the bucket has object versioning enabled. - **Copy Object** Copies an object server-side to another name or bucket, fast at any size because the data never leaves Google. Combine with Delete Object to move or rename. - **Update Object Metadata** Updates an object's content type, Cache-Control header, and custom metadata without touching its data. ### Signed URLs 1 - **Generate Signed URL** Creates a V4 signed URL that grants time-limited GET or PUT access to a single object without Google credentials, up to a 7-day expiration. This is the action that routes through human review when the link reaches an external recipient. ## Frequently Asked Questions ### What can FlowRunner do with Google Cloud Storage? FlowRunner agents can run List Buckets, Get Bucket, and Create Bucket in Google Cloud Storage, plus 9 more actions. ### Does connecting Google Cloud Storage to FlowRunner require OAuth? Google Cloud Storage uses a custom authentication method to connect to FlowRunner. ### Can Google Cloud Storage trigger a FlowRunner workflow automatically? Google Cloud Storage doesn't currently expose triggers in FlowRunner. It connects as an action step inside workflows started by another trigger. **Work at Google Cloud Storage?** This integration exposes Google Cloud Storage to AI agents on every FlowRunner plan, including through MCP, at no cost to you. [See what FlowRunner offers integration partners](https://flowrunner.ai/integrations/partners), including how to keep this page current. --- Markdown version of https://flowrunner.ai/integrations/google-cloud-storage. Site index: https://flowrunner.ai/llms.txt